Fixing a Tandy TRS-80 Model 100

The Tandy Model 100 is one of the very first notebook computers. It is a Japanese machine, designed by Kyocera  and licensed to Radio Shack in the US, and Olivetti in Europe. This notebook had great success in the early 80's among journalists and reporters. It has a great keyboard, integrated LCD display, a text editor, and easy connection to a modem. I can imagine this being the machine journalists used while attending Reagan era White House briefings.

I got mine out of ebay, the previous owner described it as not working, but mentioned that the battery LED flickered on power off, so the machine was not completely dead. Because it was listed as broken, I got a good deal. I though that if I am unable to fix it, I may put an raspberry or arduino inside.

Once I received the machine, I saw that it was in good shape cosmetically. And, yes, it was broken just as described. I inserted some new batteries (4 regular AA), and it did not even turned on. Reading online, I found that the memory switch under it must be set to ON. Then I got the flickering LED when turning off. Banging on the keyboard, I once got some beeps, so it is not dead. I saw the same behavior using an external 5V power supply (be careful that the connector is center-negative unlike most currently available power bricks!).

The Tandy model 100 has an amazing technical service manual. I also found Josh Malone's excellent document that describes some of the most common failures.

For a tear down, check out this great video by the EEVLOG:

I pressed /reset/ then /enter/ to get into BASIC, then typed "beep" + /enter/, and I got a beeping sound. This indicates that the computer is good enough to run BASIC, but the display is not working.

From Josh Malone's doc it seems that a common cause of a bad display is lack of -5V (VEE) to the LCD.

Screen and Keyboard PCBs. It is interesting how the level of integration is much higher in Japanese designed machines of this era compared to European and American designs. Each of the 10 display controller chips is responsible for a subarea of the display. They are connected mostly in parallel, and to make routing feasible using a 2 layer board, a cool trick was used: The chips in the top row are upside down and connect through holes in the PCB. Also note the high build quality, for example all screws mounts have threaded metal inserts, and machine screws are used. Most machines of the period use screws that bite directly into the plastic.

View of the main board. Note all the CMOS 4xxx discrete logic. The top ribbon cable connects to the display board, the bottom ones connect to the keyboard.

I opened the computer, and the multi-meter did not register the -5V. This confirms problems in the power circuit. Technobly describes a similar display fault that was caused by capacitor leakage. I inspected my board closely, and confirmed that many capacitors, including some used in the -5V circuit, have leaked and there is corrosion around then and under the solder mask.

I scraped the solder mask, and used white vinegar to neutralize the corrosion. Then I cleaner the vinegar with alcohol. Finally, I scrapped the visible corrosion. Luckily most corrosion was on ground planes and corroded traces seemed intact, yet it was hard to tell because the board is packed tight.

I proceeded to replace all bipolar electrolytic capacitors, even ones that looked fine. This is good practice because even good looking capacitors may have dried out after more than 35-years and be out of spec. There are a few electrolytic capacitors that seem to be unipolar. I left them, because I don't have them in my parts bin and they have not leaked yet. I need to order some and replace them at some later time.
There is also some corrosion starting on the memory backup battery, but is has not reached the terminals. I just cleaned it for now, as it still holds a charge, but I'll need to replace it soon.

Capacitors replaced near the battery positive terminal. This shows the extent of corrosion after some scrapping. On the bottom right there is a capacitor (C85) with solder mask still covering the corrosion. I later found that he trace between the smaller replaced capacitor and the resistor, just north of C100, is broken somewhere in this area. The goo under the red/black wires south of M27 is leftover flux, not corrosion. 

Corrosion under the solder mask around C85 and C84.

Capacitors and corrosion on the left side of the board. The blue capacitor seems to be unipolar.

Another view of the left side of the board.

After replacing all bipolar electrolytic capacitors I got -5V in the VEE rail, but the screen still did not turn on. I verified that I can go to Basic, and run "beep". Good, the computer was not destroyed in the recapping process.

Maybe there is also a problem with the potentiometer that controls LCD contrast. By moving the pot, I should see voltage variations in pin 4 of the display connector. But I measured no voltage there. This pin connects directly to the emitter of transistor T23. I measured it and voltage changes from 0-~4V as expected when moving the pot. I then verified that there was no continuity between the transistor and pin 4 of the display connector. I followed the trace and it goes through one of the corroded areas. I cannot see exactly where it breaks, so I did the safest thing and added a botch wire between the transistor's emitter and pin 4 on the bottom side of the motherboard.

Botch wire connecting the emitter of T23 to pin-4 of the display connector.

I reconnected the display, and it works!

It is alive! Legend says that this was the last machine that Bill Gates actively wrote code for.

It is possible that there are still corrosion problems in the machine. Some of the corrosion is in areas related to the serial I/O that I did not test yet. Also, I need to do a better job removing corrosion, for this I might have to temporarily remove a significant number of components. I'll also need to cover the exposed copper to prevent it from oxidizing.